GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling large-scale reasoning, coding, and multimodal workflows within a single system.
提供商
不同公司托管同一模型。OpenRouter 根据您选择的路由模式将请求路由到其中一家公司:Balanced(价格 + 速度)、Nitro(速度最快)或 Floor(价格最低)。
仅限 BYOK | US$2.50 | US$15.00 | US$0.25 | -- | -- |
|---|
- 输入 US$2.50 / 输出 US$15.00-- tok/s-- 正常运行时间仅限 BYOK
定价
客户为此模型实际支付的平均价格,与提供商公布的价格并列展示。由于缓存和折扣,实际支付的价格往往远低于标价。
加权平均值
GPT-5.5 (batch) — 价格历史
| US$0.00 | US$0.00 | 0.0% | 100.0% |
- 有效输入价格 US$0.00 / 实际输出 US$0.00token 占比100.0%
性能
吞吐量是模型生成内容的速度(每秒 token 数,越高越好)。延迟是完整的往返时间(越低越好)。TTFT 是首个 token 的响应时间,即您看到内容出现前需要等待的时间(越低越好)。
正常运行时间
正常运行时间是指过去 3 天中至少有一家提供商响应请求的时间百分比。可用性是指成功提供推理服务的时间百分比。OpenRouter 会持续监控,并在某家提供商返回错误时使用次优提供商。
正常运行时间数据尚不足,暂时无法显示。
当上游提供商发生错误时,如果您的请求筛选条件允许,我们可以通过路由到另一家健康的提供商来恢复服务。您可以通过以下接口以编程方式访问各提供商的正常运行时间数据: 端点 API。了解更多有关我们的 负载均衡和提供商路由选项。
基准测试
标准化评测得分,以及此模型针对共用基准测试提示词的实际输出。百分比越高越好,每项排名显示此模型在该来源所评测的模型中的位置。
| 来源 | 基准测试 | 得分 |
|---|---|---|
| Artificial Analysis | GPT-5.5 (Xhigh) Coding Index | 74.9 |
| Artificial Analysis | GPT-5.5 (Xhigh) Agentic Index | 36.4 |
| Artificial Analysis | GPT-5.5 (Xhigh) GPQA Diamond | 93.5% |
| Artificial Analysis | GPT-5.5 (Xhigh) HLE | 45.8% |
| Artificial Analysis | GPT-5.5 (Xhigh) IFBench | 75.9% |
| Artificial Analysis | GPT-5.5 (Xhigh) τ²-Bench Telecom | 93.9% |
| Artificial Analysis | GPT-5.5 (Xhigh) AA-LCR | 84.3% |
| Artificial Analysis | GPT-5.5 (Xhigh) τ-Bench Banking | 39.0% |
| Artificial Analysis | GPT-5.5 (Xhigh) GDPval-AA | 42.9% |
| Artificial Analysis | GPT-5.5 (Xhigh) CritPt | 27.1% |
| Artificial Analysis | GPT-5.5 (Xhigh) SciCode | 55.8% |
| Artificial Analysis | GPT-5.5 (Xhigh) Terminal-Bench Hard | 60.6% |
| Artificial Analysis | GPT-5.5 (Xhigh) Terminal-Bench 4.0 | 14.6% |
| Artificial Analysis | GPT-5.5 (Xhigh) Terminal-Bench 2.1 | 84.3% |
| Artificial Analysis | GPT-5.5 (Low) Coding Index | 60.9 |
| Artificial Analysis | GPT-5.5 (Low) GPQA Diamond | 91.0% |
| Artificial Analysis | GPT-5.5 (Low) HLE | 32.7% |
| Artificial Analysis | GPT-5.5 (Low) IFBench | 64.4% |
| Artificial Analysis | GPT-5.5 (Low) τ²-Bench Telecom | 83.9% |
| Artificial Analysis | GPT-5.5 (Low) AA-LCR | 81.0% |
| Artificial Analysis | GPT-5.5 (Low) τ-Bench Banking | 24.9% |
| Artificial Analysis | GPT-5.5 (Low) GDPval-AA | 27.7% |
| Artificial Analysis | GPT-5.5 (Low) CritPt | 8.0% |
| Artificial Analysis | GPT-5.5 (Low) Terminal-Bench Hard | 52.3% |
| Artificial Analysis | GPT-5.5 (Low) Terminal-Bench 2.1 | 65.5% |
| Artificial Analysis | GPT-5.5 (High) Coding Index | 71.6 |
| Artificial Analysis | GPT-5.5 (High) Agentic Index | 34.3 |
| Artificial Analysis | GPT-5.5 (High) GPQA Diamond | 93.2% |
| Artificial Analysis | GPT-5.5 (High) HLE | 45.0% |
| Artificial Analysis | GPT-5.5 (High) IFBench | 71.6% |
| Artificial Analysis | GPT-5.5 (High) τ²-Bench Telecom | 93.0% |
| Artificial Analysis | GPT-5.5 (High) AA-LCR | 84.3% |
| Artificial Analysis | GPT-5.5 (High) τ-Bench Banking | 36.7% |
| Artificial Analysis | GPT-5.5 (High) GDPval-AA | 41.8% |
| Artificial Analysis | GPT-5.5 (High) CritPt | 25.4% |
| Artificial Analysis | GPT-5.5 (High) SciCode | 56.1% |
| Artificial Analysis | GPT-5.5 (High) Terminal-Bench Hard | 59.8% |
| Artificial Analysis | GPT-5.5 (High) Terminal-Bench 4.0 | 9.1% |
| Artificial Analysis | GPT-5.5 (High) Terminal-Bench 2.1 | 79.4% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) Coding Index | 56.5 |
| Artificial Analysis | GPT-5.5 (Non-reasoning) GPQA Diamond | 76.8% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) HLE | 13.7% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) IFBench | 46.1% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) τ²-Bench Telecom | 69.3% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) AA-LCR | 64.0% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) τ-Bench Banking | 14.8% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) GDPval-AA | 23.9% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) CritPt | 1.4% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) Terminal-Bench Hard | 49.2% |
| Artificial Analysis | GPT-5.5 (Non-reasoning) Terminal-Bench 2.1 | 61.0% |
| Artificial Analysis | GPT-5.5 (Medium) Coding Index | 71.5 |
| Artificial Analysis | GPT-5.5 (Medium) Agentic Index | 29.5 |
| Artificial Analysis | GPT-5.5 (Medium) GPQA Diamond | 92.6% |
| Artificial Analysis | GPT-5.5 (Medium) HLE | 42.4% |
| Artificial Analysis | GPT-5.5 (Medium) IFBench | 71.0% |
| Artificial Analysis | GPT-5.5 (Medium) τ²-Bench Telecom | 91.8% |
| Artificial Analysis | GPT-5.5 (Medium) AA-LCR | 83.0% |
| Artificial Analysis | GPT-5.5 (Medium) τ-Bench Banking | 29.9% |
| Artificial Analysis | GPT-5.5 (Medium) GDPval-AA | 37.1% |
| Artificial Analysis | GPT-5.5 (Medium) CritPt | 18.6% |
| Artificial Analysis | GPT-5.5 (Medium) SciCode | 54.5% |
| Artificial Analysis | GPT-5.5 (Medium) Terminal-Bench Hard | 57.6% |
| Artificial Analysis | GPT-5.5 (Medium) Terminal-Bench 4.0 | 5.1% |
| Artificial Analysis | GPT-5.5 (Medium) Terminal-Bench 2.1 | 80.5% |
| Design Arena | GPT-5.5 Agents Arena Agenticgamedev Elo | 1122 |
| Design Arena | GPT-5.5 Agents Arena Agentichtmlslides Elo | 1084 |
| Design Arena | GPT-5.5 Agents Arena Agenticslides Elo | 1150 |
| Design Arena | GPT-5.5 Agents Arena Agenticslides(Html) Elo | 1077 |
| Design Arena | GPT-5.5 Agents Arena Agenticslides(Python-Pptx) Elo | 1155 |
| Design Arena | GPT-5.5 Agents Arena Androidnative Elo | 1154 |
| Design Arena | GPT-5.5 Agents Arena Full Stack Elo | 1063 |
| Design Arena | GPT-5.5 Agents Arena Godotgamedev Elo | 1149 |
| Design Arena | GPT-5.5 Agents Arena Htmlslides Elo | 964 |
| Design Arena | GPT-5.5 Agents Arena Mobile Apps Elo | 1134 |
| Design Arena | GPT-5.5 Agents Arena Pptxslides Elo | 1157 |
| Design Arena | GPT-5.5 Agents Arena Python-Pptxslides Elo | 1152 |
| Design Arena | GPT-5.5 Agents Arena Python-Pptxslides Elo | 1077 |
| Design Arena | GPT-5.5 Agents Arena Webapps Elo | 1094 |
| Design Arena | GPT-5.5 Models Arena 3D Elo | 1175 |
| Design Arena | GPT-5.5 Models Arena Asciiart Elo | 1239 |
| Design Arena | GPT-5.5 Models Arena Code Categories Elo | 1249 |
| Design Arena | GPT-5.5 Models Arena Data Visualization Elo | 1259 |
| Design Arena | GPT-5.5 Models Arena Game Development Elo | 1281 |
| Design Arena | GPT-5.5 Models Arena SVG Elo | 1224 |
| Design Arena | GPT-5.5 Models Arena UI Component Elo | 1250 |
| Design Arena | GPT-5.5 Models Arena Website Elo | 1248 |
| Vals AI | GPT 5.5 CaseLaw v2 | 66.2% |
| Vals AI | GPT 5.5 Code Migration | 45.2% |
| Vals AI | GPT 5.5 CorpFin v2 | 68.4% |
| Vals AI | GPT 5.5 EMB | 64.5% |
| Vals AI | GPT 5.5 Finance Agent (v2) | 51.8% |
| Vals AI | GPT 5.5 GPQA Diamond | 93.2% |
| Vals AI | GPT 5.5 Harvey's Legal Agent Benchmark v1 | 3.8% |
| Vals AI | GPT 5.5 LiveCodeBench | 85.3% |
| Vals AI | GPT 5.5 LegalBench | 86.5% |
| Vals AI | GPT 5.5 Legal Research Bench | 40.4% |
| Vals AI | GPT 5.5 MedCode | 49.1% |
| Vals AI | GPT 5.5 MedScribe | 86.9% |
| Vals AI | GPT 5.5 MMLU Pro | 88.1% |
| Vals AI | GPT 5.5 MMMU Pro | 88.3% |
| Vals AI | GPT 5.5 MortgageTax | 68.8% |
| Vals AI | GPT 5.5 ProgramBench | 0.5% |
| Vals AI | GPT 5.5 ProofBench v1 | 50.0% |
| Vals AI | GPT 5.5 Public Benefits Bench v1.1 | 60.9% |
| Vals AI | GPT 5.5 Vals RSI Index | 16.5% |
| Vals AI | GPT 5.5 SAGE v1 | 51.5% |
| Vals AI | GPT 5.5 SkillsBench | 62.2% |
| Vals AI | GPT 5.5 SRE Bench | 3.8% |
| Vals AI | GPT 5.5 SWE-bench | 82.6% |
| Vals AI | GPT 5.5 Tax Agent Bench | 21.8% |
| Vals AI | GPT 5.5 TaxEval v2 | 75.0% |
| Vals AI | GPT 5.5 Terminal-Bench 2.1 | 76.4% |
| Vals AI | GPT 5.5 Vals Index | 57.4% |
| Vals AI | GPT 5.5 Vals Multimodal Index | 68.1% |
| Vals AI | GPT 5.5 Vibe Code Bench v1.1 | 69.8% |
应用
向此模型发送最多流量的公开应用。这能很好地反映实际生产工作负载的情况,也能提示此模型最适合哪些用例。
使用此模型的应用尚无足够数据。
活动
此模型的 token 用量和请求流量随时间的变化。
API
可直接用于调用此模型的代码。OpenRouter 的 API 与 OpenAI 兼容——大多数 SDK 只需替换基础 URL 即可使用。切换模型时,唯一需要更改的是下方的模型标识符。
目前无法通过 EU 区域内路由使用。 区域内路由文档
文档
在 OpenRouter 上使用 OpenAI: GPT-5.5 (batch) 构建应用的指南、实用教程、博客文章和工具。
常见问题
GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling large-scale reasoning, coding, and multimodal workflows within a single system.
OpenRouter 上的 GPT-5.5 (batch) 文档包括 GPT-5.4 Migration Guide 实用指南和GPT-5.5 Price Increase: What It Actually Costs 博客文章。快速入门介绍了 API 密钥和您的首次请求,API 参考列出了所有参数。
GPT-5.5 (batch) 的费用为 US$2.50/百万输入 token 和 US$15.00/百万输出 token,Cache Read,价格为 US$0.25/M tokens和Web Search,价格为 US$10.00/1K calls 单独计费。
GPT-5.5 (batch) 的上下文窗口为 1,050,000 个 token。它最多支持 128,000 个补全 token。
是的。GPT-5.5 (batch) 接受 tools 和 tool_choice,以进行函数调用。 它还支持通过 response_format 中的 JSON schema 实现结构化输出。
GPT-5.5 (batch) 接受 PDF 等文件、图像和文本 作为输入,并返回 文本。
在 OpenRouter 上,GPT-5.5 (batch) 由 1 家提供商提供服务:OpenAI。
GPT-6.1 Terra Pro, GPT-6.1 Terra, GPT-6 Luna Decisions 及另外 58 个 是 OpenAI 的其他 文本 模型。
GPT-5.5 (batch) 于 2026年4月24日 发布。其知识截止日期为 2025年12月1日。